| public static CorrectOCRErrors ( String input ) : string | ||
| input | String | |
| 리턴 | string | |
public string PostProcess(string text)
{
if (text.Trim().Length == 0)
{
return(text);
}
// correct common errors caused by OCR
text = TextUtilities.CorrectOCRErrors(text);
// substitute Vietnamese letters frequently misrecognized by Tesseract 2.03
StringBuilder strB = new StringBuilder(text);
strB.Replace("êĩ-", "ết")
.Replace("ug", "ng")
.Replace("uh", "nh")
.Replace("rn", "m")
.Replace("iii", "m")
.Replace("II", "u")
.Replace("ôh", "ốn")
.Replace("âỳ", "ấy")
.Replace("u1I", "ưn")
.Replace("q1I", "qu")
.Replace("tmg", "úng")
.Replace("tm", "trư")
.Replace("Tm", "Trư")
.Replace("êf", "ết")
.Replace("rg", "ng")
.Replace("êh", "ến")
.Replace("fâ", "rầ")
;
// correct letter cases
text = TextUtilities.CorrectLetterCases(strB.ToString());
// add hook marks
// .ReplaceAll("(?i)(?<![q])(u)(?=[ơờởỡớợ]\\p{L})", "ư")
// .Replace("ưon", "ươn")
// .Replace("ưoi", "ươi");
string nfdText = text.Normalize(NormalizationForm.FormD);
nfdText = Regex.Replace(
Regex.Replace(
Regex.Replace(
Regex.Replace(
Regex.Replace(
Regex.Replace(nfdText,
"(?i)(?<![q])(u)(?=o\u031B" + TONE + "\\p{L})", "$1\u031B"), // uo+n to u+o+n
"(?i)(?<=u\u031B)(o)(?=" + TONE + "\\p{L})", "$1\u031B"), // u+on to u+o+n
"(?i)(i)" + TONE + "(?=[eioy])", "$1"), // remove mark on i followed by certain vowels
// It seems to be a bug with .NET: it should be \\b, not \\B,
// unless combining diacritical characters are not considered as words by .NET.
"(?i)(?<=[^q]" + VOWEL + "\\p{IsCombiningDiacriticalMarks}{0,2})(i)" + TONE + "\\B", "$1"), // remove mark on i preceeded by vowels w/ or w/o diacritics
"(?i)(?<=[aeo]\u0302)\u2019", "\u0301"), // ^right-single-quote to ^acute
"(?i)\u2018([aeo]\u0302)(?!\\p{IsCombiningDiacriticalMarks})", "$1\u0300") // left-single-quote+a^ to a^grave
;
return(nfdText.Normalize());
}
public string PostProcess(string text)
{
if (text.Trim().Length == 0)
{
return(text);
}
// correct common errors caused by OCR
text = TextUtilities.CorrectOCRErrors(text);
// substitute Vietnamese letters frequently misrecognized by Tesseract 2.03
StringBuilder strB = new StringBuilder(text);
strB.Replace("êĩ-", "ết")
.Replace("ug", "ng")
.Replace("uh", "nh")
.Replace("rn", "m")
.Replace("iii", "m")
.Replace("ll", "u")
.Replace("II", "u")
.Replace("ôh", "ốn")
.Replace("âỳ", "ấy")
.Replace("u1I", "ưn")
.Replace("q1I", "qu")
.Replace("tmg", "úng")
.Replace("tm", "trư")
.Replace("Tm", "Trư")
.Replace("êf", "ết")
.Replace("rg", "ng")
.Replace("êh", "ến")
.Replace("‘â", "ầ")
.Replace("fâ", "rầ")
;
// correct letter cases
text = TextUtilities.CorrectLetterCases(strB.ToString());
// add hook marks
// .ReplaceAll("(?<![qQ])(u)(?=[ơờởỡớợ]\\p{L})", "ư")
// .ReplaceAll("(?<![qQ])(U)(?=[ƠỜỞỠỚỢ]\\p{L})", "Ư")
// .Replace("ưon", "ươn")
// .Replace("ưoi", "ươi");
string nfdText = text.Normalize(NormalizationForm.FormD);
nfdText = Regex.Replace(
Regex.Replace(
Regex.Replace(
Regex.Replace(nfdText,
"(?i)(?<![qQ])(u)(?=o\u031B" + TONE + "\\p{L})", "$1\u031B"), // uo+n to u+o+n
"(?i)(?<=u\u031B)(o)(?=" + TONE + "\\p{L})", "$1\u031B"), // u+on to u+o+n
"(?i)(i)" + TONE + "(?=[eioy])", "$1"), // remove mark on i followed by certain vowels
"(?i)(?<=" + VOWEL + DOT_BELOW + TONE + MARK + ")(i)" + TONE + "\\b", "$1") // // remove mark on i preceeded by vowels
;
return(nfdText.Normalize());
}
public string PostProcess(string text)
{
if (text.Trim().Length == 0)
{
return(text);
}
// correct common errors caused by OCR
text = TextUtilities.CorrectOCRErrors(text);
// correct letter cases
text = TextUtilities.CorrectLetterCases(text);
return(text);
}
public static string PostProcess(string text, string langCode, string dangAmbigsPath, bool dangAmbigsOn)
{
if (text.Trim().Length == 0)
{
return(text);
}
// correct using external x.DangAmbigs.txt file first, if enabled
if (dangAmbigsOn)
{
StringBuilder strB = new StringBuilder(text);
// replace text based on entries read from an x.DangAmbigs.txt file
Dictionary <string, string> replaceRules = TextUtilities.LoadMap(Path.Combine(dangAmbigsPath, langCode + ".DangAmbigs.txt"));
if (replaceRules.Count == 0 && langCode.Length > 3)
{
replaceRules = TextUtilities.LoadMap(Path.Combine(dangAmbigsPath, langCode.Substring(0, 3) + ".DangAmbigs.txt")); // fall back on base
}
if (replaceRules.Count == 0)
{
throw new NotSupportedException(langCode);
}
Dictionary <string, string> .KeyCollection.Enumerator enumer = replaceRules.Keys.GetEnumerator();
while (enumer.MoveNext())
{
string key = enumer.Current;
string value = replaceRules[key];
strB = strB.Replace(key, value);
}
text = strB.ToString();
}
// postprocessor
text = PostProcess(text, langCode);
// correct common errors caused by OCR
text = TextUtilities.CorrectOCRErrors(text);
// correct letter cases
return(TextUtilities.CorrectLetterCases(text));
}